Although it takes a lot of time and effort, in the beginning, stages to create a product you imagined, the work is often well worth it when it sells well to make up for the expenses used to make it and then some. Financial success for entrepreneurs varies, depending on their marketing skills and how useful their product actually is, but most wonder how retailers actually distribute their products.
It is the same question Kris Ferreira, an assistant professor of business administration asked at a presentation at Future Assembly. Every retailer much faces tricky and tactical choices related to product placement, assortment, pricing, and inventory. Ferreira mentions that every single one of those choices would have been simple to make if retailers were made aware of consumer demands we.
She added that the primary issue for her was that she had a lot of uncertainty when it came to the demand for her product. Nevertheless, it’s a less than complicated riddle that can get solved. Ferreira believes that the trick one should use for tactical design making such as this lies in quantitative analysis. She stated that the business world uses a few interesting analytics for products. One of them is called descriptive analytics. It analyzes what has already happened. Another is called predictive analytics, which analyzes data to figure out what to do next. The third is prescriptive analytics, which uses data to choose what to do next.
These forms of analysis involve events, analytics, and products. Events include the date length, and the type of event. Analyicits consists of discount percentage, cost, relative cost of competing styles, amount of styles bought in the same subclass and event, the number of branded events in the last year, and the number of concurrent events in the department. Products include the class, color popularity, size popularity, type of brand (A or B), the popularity of the brand, and department.
Ferriera believes to be able to combine predictive analytics to predict demand with prescriptive analytics so they make tactical choices, lies in the data. She showed field work that she and her colleagues worked on along with Rue La La, an online retailer located in Boston. The Boston-based company is known to be a flash sales business that offers highly discounted and limited time offers on accessories and designer clothing. Most of the “limited time” offers that they sell in the store include items the retailer never sold at their store, and a couple of them would sell out quickly. In that event, the retailer takes it as a sign that they could have charged more for the product.
On the other side of the fence, there were products that didn’t sell well, which would signify that the products were priced too high. The primary challenge to those conducting the research was figuring out how to predict demand and create prices to maximize the revenue of the new products with no prior sales data. They decided to make a pricing decision tool that had the ability to use current data to maximize revenue on new products. The researchers took advantage of machine learning techniques that could estimate lost sales in the past (the products that sold out), and predict the demand for a new product the company would sell in the future.
While they were working on the research, they discovered that the demand for a certain product also depended on the cost of other products in the same category. From that finding, the researchers made a highly efficient multi-product price optimization algorithm to suggest a cost for all the items listed on Rue La La’s website on a given day. In January of 2015, the researchers worked alongside Rue La La during a field experiment to assist the retailer in creating optimal prices for new products they will add. The researchers were able to show how Rue La La was able to increase their revenue of products in the experiment by nearly 10% via price recommendations from the algorithm. It also had a low impact on gross sales quantity.
Ten percent may not seem like much, but it’s a big difference, especially in preventing experiencing a loss in sales. Even though the research gave its undivided attention to a flash sales setting, they are certain their techniques would also be extremely useful for just about any other retailer. It doesn’t matter if they operate fully online, at physical locations or a blend of the two. It would be beneficial to them anytime they had to create unique costs for new products they would sell before they make them available for consumers.
In a broader aspect, the work the researchers conducted positively illustrated how both prescriptive analytics and predictive analytics can get mixed to create a tactical decision-making tool that can largely impact how well sales can go for a product.